Communication: Surface Structures of Native Bacteriorhodopsin Depend on the Molecular Packing Arrangement in the Membrane

ArticleNo.jmbi.1998.2441availableonlineathttp://www.idealibrary.comon J. Mol. Biol. (1999) 285, 1903±1909

COMMUNICATION Surface Structures of Native Bacteriorhodopsin Depend on the Molecular Packing Arrangement in the Membrane

DanielJ.MuÈller1*,Hans-JuÈrgenSass2,ShirleyA.MuÈller1,GeorgBuÈldt2 andAndreasEngel1

1M. E. MuÈller-Institute for Bacteriorhodopsin is the one of the best-studied models of an ion pump. Microscopic Structural Biology Five atomic models are now available, yet their comparison reveals Biozentrum University of Basel differences of some loops connecting the seven transmembrane a-helices. Klingelbergstrasse 70 In an attempt to resolve this enigma, topographs were recorded in aqu- CH-4056 Basel, Switzerland eous solution with the atomic force microscope (AFM) to reveal the most native surface structure of bacteriorhodopsin molecules in the purple 2Forschungszentrum JuÈlich membrane. Individual peptide loops were observed with a lateral resol- IBI-2: Structural Biology ution of between 4.5 AÊ and 5.8 AÊ , and a vertical resolution of about 1 AÊ . D-52425 JuÈlich, Germany The AFM images demonstrate for the ®rst time, that the shape, the position, and the ¯exibility of individual polypeptide loops depend on the packing arrangement of bacteriorhodopsin molecules in the lipid bilayer. # 1999 Academic Press *Corresponding author Keywords: bacteriorhodopsin; purple membranes; atomic force microscopy

Bacteriorhodopsin (BR), a light-driven proton cubiclipidphase(Landau&Rosenbusch,1996) pump(Oesterhelt&Stoeckenius,1973),ispacked have recently allowed the structure of BR to be into highly ordered trigonal two-dimensional (2D) determinedtoaresolutionof2.5AÊ(Pebay- lattices(Henderson,1975)toformthepurplemem- Peyroulaetal.,1997)and2.3AÊ(Lueckeetal.,1998) brane of Halobacterium salinarum. Electron crystallo- by X-ray crystallography. Furthermore, hetero- graphy of these membranes produced the ®rst geneous nucleation of BR trimers into 3D crystals three-dimensional (3D) structure of a membrane allowed preservation of the purple membrane lipid protein(Henderson&Unwin,1975),and®nally composition and the structural determination of the®rstatomicmodelofBR(Hendersonetal., the bacteriorhodopsin-lipid complex to 2.9 AÊ 1990;Grigorieffetal.,1996).Theloopsconnecting (Essenetal.,1998).Theatomicmodelscorrelate the a-helices A and B, B and C, and E and F well in the a-helical regions, but differ signi®cantly appeared to be disordered as they could not be in their surface structures. In the X-ray maps of BR resolved reliably. Progress in instrumentation and crystals grown in a cubic lipid phase, some of the the improved structural preservation obtained by peptide loops were visualized clearly, while others trehalose embedding and rapid-freezing techniques appeared to be disordered. This suggests that the (Hendersonetal.,1990)havemadeitpossibleto experimentalmethodsemployedtosolvetheBR collect electron microscopy data to a resolution of structure in¯uenced the structure of the hydrophi- 3AÊ(Kimuraetal.,1997).Asaresult,mostloops lic loops at the BR surfaces. Hence it would be have now been resolved as clearly as the a-helical meaningful to image BR in its most native state transmembrane regions. 3D BR crystals grown in a (Sassetal.,1997),namelyinthemembraneandin buffer solution. Here, we compare atomic force microscopy Abbreviations used: AFM, atomic force microscopy; (AFM) topographs of BR in native purple mem- BR, bacteriorhodopsin; RMS, root-mean-square; SD, standard deviation; 2D, two-dimensional; 3D, three- branes(MuÈlleretal.,1995a,b,1996)andofBR dimensional. assembled in vitro into an orthorhombic lattice E-mail address of the corresponding author: (Micheletal.,1980).Alltopographswererecorded [email protected] in the same buffer solution (100 mm KCl, 10 mM

0022-2836/99/051903±07 $30.00/0 # 1999 Academic Press 1904 Surface Structures of Native Bacteriorhodopsin

Tris(pH 7.8), and under identical scan conditions, E-F loops were bent away and the shorter loops of except that different forces were applied to the theBRmonomerswerevisualized(Figure1(c)). stylus. This conformational change is fully reversible (MuÈlleretal.,1995b),suggestingthatloopE-Fisa High-resolution topographs were acquired rather ¯exible element on the cytoplasmic side of in buffer solution the BR molecule. At this force of 200 pN, the maximum height difference between the protein and The cytoplasmic BR surface imaged with a force the lipid membrane was 6.4(Æ1.2) AÊ (n 398). of 100 pN applied to the AFM stylus revealed tri- Three distinct protrusions were recognized in meric structures arranged in a trigonal lattice of almost every monomer, and a further distinct pro- 62(Æ2)AÊsidelength(Figure1(a)).Subunitsinthe trusion was present at the center of the trimers. trimer featured a particularly pronounced protru- Thecalculateddiffractionpattern(Figure1(d)) sion extending 8.3(Æ1.9) AÊ (n 398) above the documents an isotropic resolution out to 4.5 AÊ . lipid surface. As shown previously, this protrusion Topographs of the native extracellular BR sur- arises from the loop connecting transmembrane face acquired with the AFM using an applied force a-helicesEandF(MuÈlleretal.,1995b).Thepower of 100 pN revealed the arrangement of tripartite spectrum of the topograph included 11th-order dif- protrusionsonatrigonallattice(Figure1(e)).The fractionspots(Figure1(b))demonstratingthata maximum height difference between the protein lateral resolution of 4.9 AÊ has been achieved. and the lipid membrane was 5.3(Æ0.7) AÊ (n 320). At applied forces above 200 pN, AFM topo- Thepowerspectrum(Figure1(f))exhibitedcharac- graphs were signi®cantly different. The prominent teristic strong second-order spots, and extended to

Figure 1. Purple membrane surfaces as observed in buffer solution using the AFM. (a) Native cytoplasmic surface recorded at 100 pN. (b) Power spectrum of (a). (c) Cytoplasmic surface recorded at 200 pN. (d). Power spectrum of (c). (e) Native extracellular surface recorded at 100 pN. (f) Power spectrum of (e). (g) Orthorhombic crystal of BR recorded at 100 pN. In this crystal form (p22121) the rows of BR dimers alternate, to expose either their cytoplasmic or their extracellular surfaces to the aqueous solution. (h) Power spectrum of (g). Vertical brightness range of topographs: 10 AÊ . Purple membranes were isolated from Halobacterium salinarum (ET1001) (Oesterhelt & Stoeckenius, 1974) and orthorhombic 2D crystals were prepared as described (Michel et al., 1980). In all cases, the samples were adsorbed onto freshly cleaved mica in buffer solution (300 mM KCl, 10 mM Tris-HCl (pH 7.8); MuÈ ller et al., 1997), while the buffer solution used for recording the high-resolution topographs was 100 mM KCl, 10 mm Tris-HCl (pH 7.8). The atomic force microscope used was a Nanoscope III (Digital Instruments, Santa Barbara, California) equipped with an E-scanner (12 mm) and oxide sharpened Si3N4 tips on a cantilever with a spring constant of 0.1 N/ m (Olympus, Tokyo, Japan). Topographs recorded in trace and retrace scanning directions using contact mode showed no signi®cant differences. All measurements were carried out at room temperature. Surface Structures of Native Bacteriorhodopsin 1905 the 11th order indicating a lateral resolution of 4.9 AÊ . In contrast to the cytoplasmic surface, increasing the force applied to the stylus did not cause structural changes in the individual extracellular loops. However, in these images it is dif®- cult to assign the center of the BR trimer unambiguously(MuÈlleretal.,1996).Forthis reason, we also investigated in vitro reassembled membranes of a dimeric form of BR. Recrystallization of BR in the presence of n-dodecyl trimethylammonium chloride (DTAC) yieldedwell-ordered2Dcrystals(Micheletal1980) that adsorbed ¯atly onto freshly cleaved mica. They had sizes of up to 5 mm, and a thickness of 58(Æ4) AÊ (n 113), which was slightly more than thatofthenativepurplemembrane55(Æ4)AÊ (MuÈller&Engel,1997).Asexpected,topographsof the orthorhombic crystal showed BR dimers assembled into a rectangular lattice with a p22121 symmetryandunitcelldimensionsofa58AÊ, b74AÊ(Figure1(g));Micheletal.,1980).Accord- ingly, the BR dimers alternately had their cytoplasmic surface or their extracellular surface facing the stylus. The maximum height difference between the protrusions and the bilayer was Figure 2. Averaged cytoplasmic surface of the BR tri- 8.1(Æ0.9) AÊ (n 368). The power spectrum mer of purple membrane. (a) Native cytoplasmic surface revealed diffraction spots to the tenth order recorded at applied forces of 100 pN (average of 398 indicatingaresolutionof5.8AÊ(Figure1(h)). unit cells). (b) Native cytoplasmic surface recorded at Surprisingly, it was not possible to induce confor- applied forces of 200 pN (average of 380 unit cells). Cor- mational changes of the E-F loops in this BR crystal relation averages are displayed in perspective view (top, shaded in yellow-brown) and in top view (bottom, in form. Increasing the applied force of the stylus blue) with full brightness ranges of 10 AÊ . To assess the resulted in a deformation of the whole protein sur- ¯exibility of the different structures, standard deviation face rather than in the bending of a single loop, (SD)mapswerecalculated(Karraschetal.,1994;MuÈller and reduced the lateral resolution. etal.,1998),andhadarangefrom0.8(lipid)to1.9AÊ A striking feature of all unprocessed topo- (extendedE-Floop;seeTable1).Surfaceregionsexhibit- graphsdisplayedinFigure1istheirhighsignal- ing a SD above 1.2 AÊ are superimposed in red-to-white to-noise ratio. Smallest details are not only shades. The outlined BR shapes were adapted from sec- resolved laterally, but their height is highly tions close to the cytoplasmic surface of BR trimers reproducible as revealed by comparing different obtained from electron crystallographic analyses unit cells and considering that the full gray level (Figure3;Grigorieffetal.,1996).Thecorrelation-aver- Ê aged topographs were 3-fold symmetrized and exhibited range represents 10 A. 9.2 % (a), and 14.1 % (b) root-mean-square (RMS) deviation from P3 symmetry. Averaged AFM topographs reveal structural changes on the cytoplasmic surface of BR Imaged in its most extended form the E-F loop of bacteriorhodopsin from the native purple mem- bilayer by approximately 5 AÊ . Nevertheless, helix F brane reached its maximum height above the lipid represents the highest protrusion at the cyto- bilayerof8.3AÊclosetohelixF(Figure2(a)).While plasmic surface in this structure as well. Two the standard deviation (SD) of the height measure- minor protrusions in the AFM topograph are loca- ments was around 1 AÊ for most morphological fea- lized close to helix B and between helices C and D, tures of the topography, the E-F loop exhibited a respectively. When the major protrusion represent- SDof1.9AÊ(Table1),consistentwiththehightem- ing loop E-F had been pushed away by applying a perature factor observed by electron microscopy force of 200 pN to the stylus, the cytoplasmic sur- (Grigorieffetal.,1996).Theheightof8.3AÊcorre- face of the BR molecule appeared different and lates well with the 3D maps from electron crystal- exhibited®nerdetails(Figure2(b)).Theprotrusion lographicanalyses(Figure3)whichshowedhelixF between helices F and G (no. 4) together with the to protrude out of the bilayer by 7 to 10 AÊ , minor elevation close to helix E (no. 3) is likely to whereas helix E either did not extend into the aqu- represent what remained from loop E-F and the eousphase(Kimuraetal.,1997),orprotrudedonly protruding parts of helices E and F that are com- slightlyabovethebilayer(Figure3(a);Grigorieff pressed by the AFM stylus. However, it cannot be etal.,1996)).However,intheX-raystructureby excluded that the protrusion between helices F and Essenetal.(1998)helixEprotrudesoutofthe G represents part of the 25 amino acid residue long 1906 Surface Structures of Native Bacteriorhodopsin

Table 1. Heights (in AÊ ) of protrusions in the trigonal and orthorhombic crystals

Heights (AÊ ) Æ SD of: Trigonal crystal Orthorhombic crystal Membrane 55 Æ 458Æ 4

Cytoplasmic surface: (n 398) (n 368) Extended E-F loop 8.3 Æ 1.9 - Protrusion 1 5.8 Æ 0.8 4.5 Æ 0.9 Protrusion 2 5.6 Æ 0.7 6.8 Æ 0.9 Protrusion 3 4.2 Æ 1.1 6.8 Æ 0.9 Protrusion 4 6.4 Æ 1.2 8.1 Æ 0.9 Protrusion 5 3.6 Æ 0.9 3.9 Æ 09 Protrusion 6 - 7.6 Æ 0.9

Extracellular surface: (n 320) (n 368) Protrusion 7 3.3 Æ 1.2 2.2 Æ 0.7 Protrusion 8 2.7 Æ 0.7 4.0 Æ 0.8 Figure 3. Structure of bacteriorhodopsin. (a) Ribbon Protrusion 9 5.3 Æ 0.7 3.8 Æ 0.9 representation as revealed by electron crystallography Protrusion 10 - 2.8 Æ 0.6 (Grigorieffetal.,1996).Becauseoftheirdisordering,the Protrusion 11 3.3 Æ 0.7 - N terminus of helix A and the C terminus of helix G Protrusion 12 0.8 Æ 0.7 - Protrusion 13 1.4 Æ 0.7 - were not resolved and the B-C loop was only partly resolved. Blue lines indicate the cytoplasmic and extra- The numbers of protrusions refer to protrusions de®ned in cellularsurfacesdeterminedbyKimuraetal.(1997). theaveragedtopographs(Figures2and4).Verticaldistances (b) Outlines of 10 AÊ thick slices of the cytoplasmic and are given relative to the surface of the lipid bilayer, except for the extracellular BR surface. (Data were kindly provided the membrane thicknesses which were measured relative to the support and in the absence of repulsive electrostatic double- byN.GrigorieffandR.Henderson(Grigorieffetal., layerinteractions(MuÈller&Engel,1997).Thestandarddevia- 1996).) tions (SD) of the averaged topographs were calculated during the averaging of the unit cells. The number of unit cells averaged is represented by n independentoftheappliedforce(Figure4(a);protrusions no. 3 and no. 6). The triangular protrusion C-terminal domain. This uncertainty arises because (no. 4) located between helices B and G may result the AFM height signal in this area exhibited a from the C terminus. None of these structures signi®cant standard deviation (red shaded in exhibited signi®cant variabilities, indicating a Figure2(b)andseeTable1),consistentwiththe structural stabilization by the different packing elevated temperature factor of this region deter- arrangement in the orthorhombic compared to minedbyelectronmicroscopy(Grigorieffetal., the trigonal lattice. An additional protrusion 1996).TheotherprotrusionsintheAFMtopograph (Figure4(a),no.5)wasobservedattheperiphery may be assigned by comparison with the atomic of each BR monomer packed in the orthorhombic modelsderivedfromtheBRtrimer(Grigorieffetal., lattice, probably representing bound lipid mol- 1996;Kimuraetal.,1997;Essenetal.,1998).In ecules(Grigorieffetal.,1996). these models helix B protrudes out of the bilayer, The observed structural changes suggest that the and helix A ends below the bilayer surface interactions of the cytoplasmic polypeptide loops (Figure3(a)).Therefore,thetopographicprotru- depend on how the BR molecules associate. In the sion no. 1 is likely to represent the short loop BR trimer, there is a crevice between helices A and (seven amino acid residues) connecting helices A B, and helices E and D of neighboring monomers andB(Figure2(b)).Inaddition,thediscretepro- (Figure2;outlines).Lipidmoleculesinthiscrevice trusion between helices C and D (no. 2) correlates arestable(Grigorieffetal.,1996),andstabilizethe to their connecting loop (three amino acid resi- BR trimer by speci®c interactions with their lipid dues). A further protrusion (no. 5) of 2 AÊ height andhead-groupmoieties(Essenetal.,1998).This was present at the 3-fold axis of the BR trimer crevice is not present in the orthorhombic BR and probably arises from lipid molecules assembly and hence, the different molecular inter- (Grigorieffetal.,1996). actions probably allow the displacement of the The arrangement of the protrusions on the cyto- loopconnectinghelicesAandB(Figure4(a);white plasmic face of BR was very distinct when AFM contours). Helices F and G of two neighboring BR topographs of the orthorhombic in vitro assemblies molecules are closely packed providing space for were analysed. The protrusion of the A-B loop (no. at most two lipid molecules. Consistent with the 1) was shifted by 3 AÊ , now being located between stability of the E-F loops the standard deviation thepositionofhelicesAandB(Figure4(a)).The over this region was not enhanced as in the trigo- short loop (three amino acid residues) connecting nalcrystal(Table1).DifferencesinhelixEhave helices C and D was observed as a discrete protru- also been observed in X-ray structures from differ- sion (no. 2) in the orthorhombic lattice, close to its ent crystal forms. While the end of helix E has not position in the trigonal lattice. Remarkably, the E-F been resolved in the crystals grown in the cubic loop was observed as a bean-shaped structure lipidphase(Pebay-Peyroulaetal.,1997;Luecke Surface Structures of Native Bacteriorhodopsin 1907

orthorhombic lattice, away from the intermolecular space.

Extracellular surface features differ between BR crystal forms The extracellular surface of BR assembled into anorthorhombiclattice(Figure4(a);yellowout- line) exhibited distinct features which were corre- lated to the atomic model. Protrusion no. 7 of the monomer was located close to helix A and may also represent part of the N-terminal domain (seven amino acid residues). Whereas no protrusion was observed at the position of helix B, protrusions no. 8 and no. 9 correlate with the b-hairpinconnectinghelicesBandC(Kimuraetal., 1997;Essenetal.,1998).Theshortloopconnecting helices E and D appears to be represented by protrusion no. 10. The region between helices F and G exhibited an enhanced standard deviation (1.7 AÊ ; Figure4(a);redcolored)andtheconnectingloop did not make a structural contribution to the average. Thus, in contrast to the cytoplasmic surface, Figure 4. Averaged surface of BR assembled into an orthorhombic lattice (a) and of the extracellular surface the crystal contacts along helices F and G did not of purple membrane (b) Correlation averages (average stabilize the corresponding extracellular loops. of 368 unit cells (a) and of 320 unit cells (b)) were dis- On the extracellular surface of native purple played in perspective view (top, shaded in yellow- membrane(Figure4(b),protrusionno.7arising brown) and in top view (bottom, shaded in blue) with a from the N-terminal region is at the same position vertical brightness range of 10 AÊ . The SD map had a in the BR trimer as in the orthorhombic crystal. In range from 0.7 (lipid) to 1.7 AÊ (F-G region). Surface contrast to the latter, a small protrusion (no. 11) is regions exhibiting a SD above 1.2 AÊ are superimposed observed at the location of helix B. The b-hairpin inred-to-whiteshades(seeTable1).Thecorrelation- (Kimuraetal.,1997;Essenetal.,1998),represented averaged topograph of the orthorhombic lattice (a) was by protrusions no. 8 and no. 9, starts at helix B, 2-fold symmetrized exhibiting a 2.1 % RMS deviation from P2 symmetry, whereas the 3-fold symmetrized propagates back into the lipid bilayer to create a extracellular surface (b) exhibited an 6.1 % RMS devi- depression seen between protrusion no. 11 and no. ation from P3 symmetry. Topographs were recorded at 8. Compared to the topography of the orthorhom- applied forces of 100 pN. The outlined sections were bic crystal, protrusion no. 9 is shifted by 11 AÊ adapted from electron crystallographic analyses. Based away from the center of the BR molecule and its on the projection map of orthorhombic BR crystals height above the bilayer is increased by 1 AÊ . The (Micheletal.,1980;Leifer&Henderson,1983),cyto- rather large rearrangement of the B-C loop plasmic and extracellular sections of the BR molecule observed in the orthorhombic lattice presumably revealedfromtrigonalcrystals(Grigorieffetal.,1996) results from the different crystal contacts and the were arranged on the orthorhombic lattice. With the altered proteinaceous and lipidic interactions. Simi- knowledge of the extracellular protrusions on the orthorhombiccrystal(Figure4(a)),itwaspossibleto larly, the loop connecting helices F and G, which is align the averaged AFM topographs of the extracellular mobile in the orthorhombic lattice, was observed bacteriorhodopsinsurface(Figure4(b))withtheextra- as a stable protrusion (no. 12) in the BR trimer and cellular section of the BR trimers obtained from elec- exhibited no signi®cant standard deviation. Protru- troncrystallography(Figure3(b));Grigorieffetal., sionno.13(Figure4(b))wasobservedatthepos- 1996). ition of the triglycoside head-groups of lipids which were found to occur ordered in the crevice betweentwoBRmonomers(Essenetal.,1998).In addition, the topograph of native purple membrane shows a depression at the center of the extra- cellularsideoftheBRmolecule(Figure4(b),where etal.,1998),helixEwasstableandfullyresolved protons are assumed to exit the protein. inthestructurebyEssenetal.(1998),wherecrystal contacts along helices F and G occur. Accordingly, Conclusions we conclude that in the orthorhombic BR crystals the interactions between helices F and G of two As demonstrated by comparing the atomic adjacent BR molecules affected both the structural modelsofBRpresentedbyGrigorieffetal.(1996), appearance and the rigidity of the E-F loop and of Kimuraetal.(1997),Pebay-Peyroulaetal.(1997), the C-terminal region. In addition, the protrusion Essenetal.(1998)andLueckeetal.(1998),struc- of loop A-B was shifted towards helix A in the tures which are shielded by the hydrophobic belt 1908 Surface Structures of Native Bacteriorhodopsin of the bilayer are not greatly in¯uenced by the Hargrave, P. A. (1991). Seven-helix receptors. Curr. preparation technique. This is in contrast to struc- Opin. Struct. Biol. 1, 575-581. tures that face the aqueous solution. The presented Heberle, L., Riesle, L., Thiedemann, G., Oesterhelt, D. & AFM topographs of two BR crystal forms clearly Dencher, N. A. (1994). Proton migration along the show differences in structural detail re¯ecting the membrane surface and retarded surface to bulk shifting of loops of the BR molecule. Interestingly, transfer. Nature, 370, 379-382. Henderson, R. (1975). The structure of the purple mem- the ¯exibility of some of the loops was also in¯u- brane from Halobacterium halobium: analysis of the enced. Since the sensitivity (2) of the AFM is suf®- X-ray diffraction pattern. J. Mol. Biol. 93, 123-138. cient to image individual loops in their extended Henderson, R. & Unwin, P. N. T. (1975). Three-dimen- conformation(MuÈ1leretal.,1995b),thedefor- sional model of purple membrane obtained by elec- mations induced by the imaging process at forces tron microscopy. Nature, 257, 28-32. of <100 pN can be neglected. However, it was Henderson, R., Baldwin, J. M., Ceska, T. A., Zemlin, F., possible to force loop E-F of the BR trimer to Beckman, E. & Downing, K. H. (1990). Model for undergo a conformational change of about 2 AÊ by the structure of bacteriorhodopsin based on high- increasing the force applied to the AFM cantilever resolution electron cryo-microscopy. J. Mol. Biol. >200 pN, which corresponds to an energy differ- 213, 899-929. ence of 4 kJ/mol. Our experimental results, there- Karrasch, S., Hegerl, R., Hoh, J., Baumeister, W. & fore, demonstrate for the ®rst time that the shape, Engel, A. (1994). Atomic force microscopy produces the position, and the ¯exibility of individual poly- faithful high-resolution images of protein surfaces in an aqueous environment. Proc. Natl Acad. Sci. peptide loops exposed to the aqueous solution USA, 91, 836-838. depend on the packing arrangement of proteins Kimura, Y., Vassylev, D. G., Miyazawa, A., Kidera, A., within the lipid bilayer. Matsushima, M., Mitsuoka, K., Murata, K., Hirai, T. Although such structural changes may not sig- & Fujiyoshi, Y. (1997). Surface of bacteriorhodopsin ni®cantly in¯uence the proton pumping function revealed by high-resolution electron microscopy. ofbacteriorhodopsin(Micheletal.,1980),theymay Nature, 389, 206-211. in¯uencethepathwayofprotonsontheprotein Landau, E. M. & Rosenbusch, J. P. (1996). Lipidic cubic surface(Heberleetal.,1994),atopicwhichispre- phases: a novel concept for the crystallization of sently discussed on the basis of static models membrane proteins. Proc. Natl Acad. Sci. USA, 93, (Kimuraetal.,1997;Essenetal.,1998;Lueckeetal., 14532-14535. 1998).Bacteriorhodopsin,oneofthebest-character- Leifer, D. & Henderson, R. (1983). Three-dimensional structure of orthorhombic purple membrane at ized transmembrane proteins, exhibits structural Ê homologies to the seven-helix G-protein-coupled 6.5 A resolution. J. Mol. Biol. 163, 451-466. Luecke, H., Richter, H.-T. & Lanyi, J. K. (1998). Proton receptorfamily(Hargrave,1991;Baldwin,1993). transfer pathways in bacteriorhodopsin at 2.3 AÊ ng- The observed structural changes may be of crucial strom resolution. Science, 80, 1934-1937. importance for these membrane proteins whose Michel, H., Oesterhelt, D. & Henderson, R. (1980). surface regions are known to interact with other Orthorhombic two-dimensional crystal form of pur- proteins. ple membrane. Proc. Natl Acad. Sci. USA, 77, 338- 342. MuÈ ller, D. J. & Engel, A. (1997). The height of biomole- cules measured with the atomic force microscope depends on electrostatic interactions. Biophys. J. 73, Acknowledgments 633-1644. MuÈ ller, D. J., Schabert, F. A., BuÈ ldt, G. & Engel, A. We thank Jacques Dubochet, Joerg Kistler, Ehud (1995a). Imaging purple membranes in aqueous sol- Landau, and Jurg Rosenbusch for critical reading of the utions at subnanometer resolution by atomic force manuscript. This work was supported by the Swiss microscopy. Biophys. J. 68, 1681-1686. National Foundation for Scienti®c Research, grant MuÈ ller, D. J., BuÈ ldt, G. & Engel, A. (1995b). Force- 31-424335.94 to A.E., and the Maurice E. MuÈ ller Foun- induced conformational change of bacteriorhodop- dation of Switzerland. sin. J. Mol. Biol. 249, 239-243. MuÈ ller, D. J., Schoenenberger, C. A., BuÈ ildt, G. & Engel, A. (1996). Immuno-atomic force microscopy of purple membrane. Biophys. J. 70, 1796-1802. References MuÈ ller, D. J., Amrein, M. & Engel, A. (1997). Adsorption Baldwin, J. M. (1993). The probable arrangement of the of biological molecules to a solid support for scan- helices in G protein-coupled receptors. EMBO J. 12, ning probe microscopy. J. Struct. Biol. 119, 172-188. 1693-1703. MuÈ ller, D. L., Fotiadis, D. & Engel, A. (1998). Mapping Essen, L.-O., Siegert, R., Lelimann, W. D. & Oesterhelt, ¯exible protein domains at subnanometer resolution D. (1998). Lipid patches in membrane protein oligo- with the AFM. FEBS Letters, 430, 105-111. mers: Crystal structure of the bacteriorhodopsin- Oesterhelt, D. & Stoeckenius, W. (1973). Functions of a lipid complex. Proc. Natl Acad. Sci. USA, 95, 11673- new photoreceptor membrane. Proc. Natl Acad. Sci. 11678. USA, 70, 2853-2857. Grigorieff, N., Ceska, T. A., Downing, K. H., Baldwin, Oesterhelt, D. & Stoeckenius, W. (1974). Isolation of the J. M. & Henderson, R. (1996). Electron-crystallo- cell membrane of Halobacterium halobium and its graphic re®nement of the structure of bacteriorho- fraction into red and purple membrane. Methods dopsin. J. Mol. Biol. 259, 393-421. Enzymol. 31, 667-678. Surface Structures of Native Bacteriorhodopsin 1909

Pebay-Peyroula, E., Rummel, G., Rosenbusch, J. P. & Sass, H. J., Schachowa, I. W., Rapp, G., Koch, M. H. J., Landau, E. M. (1997). X-ray structure of bacterior- Oesterhelt, D., Dencher, N. A. & BuÈ ldt, G. (1997). hodopsin at 2.5 AÊ ngstroms from microcrystals The tertiary structural changes in bacteriorhodopsin occur between M states: X-ray diffraction and Four- grown in lipidic cubic phases. Science, 277, 1676- ier transform infrared spectroscopy. EMBO J. 16, 1681. 1484-1491.

Edited by W. Baumeister

(Received 20 August 1998; received in revised form 18 November 1998; accepted 20 November 1998)